Touting the large number of high-performance computing (HPC) applications that incorporate GPU acceleration, Nvidia on Monday announced new software and tools on the Nvidia GPU Cloud (NGC) container registry that allow scientists to quickly deploy scientific computing applications and HPC visualization tools.
Typically, scientists who want to use these applications face time-consuming installation challenges and resource-intensive updates. Now, accessing these applications should be “as easy as dowloading an app from the App Store,” Ian Buck, vice president and general manager of Accelerated Computing for Nvidia, told reporters.
HPC application and HPC visualization containers are available from the NGC container registry for anyone with an NGC account. The HPC containers can run on any Nvidia Pascal and later generation Nvidia GPU-accelerated system.
Nvidia announced the tools at the SuperComputing17 conference in Denver, to demonstrate the company’s growing relevance in high-performance computing.
Already, more than 500 HPC applications are GPU-accelerated. A new report by analyst firm Interesect360 Research shows that this includes the top 15 and 70 percent of the top 50 HPC applications. From astronomy to life sicences and medical imaging, GPU-accelerated applications are used in a range of scientific fields. They are, however, simply hard to install.
“They’re built from many different software stacks and libraries,” Buck explained. “They’re built by researchers for themselves, not necessarily for releasing out into the wild — it’s just not their priority.”
Nvidia addressed the same problem in the deep learning and cloud community, when it announced the GPU Cloud for AI developers last month. The new tools are now a part of that container registry.
In terms of the HPC applications coming to the NGC, Nvidia is starting out small, with five major applications — GAMESS, GROMACS, LAMMPS, NAMD and RELION — with more on the way.
Meawnwhile, for HPC visualization, Nvidia has worked with ParaView to introduce three containers now available in beta: ParaView with Nvidia IndeX is for visualizing large-scale volumetric data; ParaView with Nvidia Optix is deployed for ray-tracing, and Nvidia Holodeck offers interactive real-time visualization and high-quality visuals.
Nvidia on Monday also hailed the popularity of the Tesla V100 GPU, based on Nvidia’s Volta architecture, which is now available through every major sever vendor and major cloud service provider to deliver artificial intelligence and high performance computing. That includes Dell EMC, HPE, Huawei, IBM and Lenovo, as well as cloud providers Alibaba Cloud, Amazon Web Services, Baidu Cloud, Microsoft Azure, Oracle Cloud and Tencent Cloud.
Previous and related coverage:
- Nvidia’s Tesla V100 GPU gets broad backing as server vendors eye AI workloads
- Nvidia’s data center business hits $2 billion annual revenue run rate